AMENDED SUBMISSION TO THE CITIZENS ASSEMBLY
AN 'AV/MIXED MEMBERSHIP PROPORTIONALITY SYSTEM'
(Original submission
Devine
1343 - August 13, 2004)
This is an amended version of previous submission (
Divine
1343). This version expands, clarifies, and improves the
original submission.Two appendices show the results of using
section II of this system in the 2001 and 1996 provincial
elections.
Thank you for the very important work you are doing. A fair,
workable voting system is the foundation of democratic government.
The many distortions in our present system deprive too many
citizens of a voice in the electoral process and discourage them
from voting.
Here are my objections to the present system:
1) It has given us a majority party with a grossly disproportional
share of the legislature. (2001)
2) It has given us a majority party that decisively lost the
popular vote. (1996)
3) Minor parties have not been represented or are have been grossly
under-represented in the legislature.
4) The ideas and views of minor parties have not been debated or
even given serious voice in the legislature.
5) We seem to oscillate from one arrogant majority party in power
to the next arrogant majority party.
6) Under the present system, majority party lawmakers too often act
as rubberstamps, mindlessly cheerleading legislation that they've
barely looked at. Debate is an endless sham of empty posturing and
self-congratulation. How can sound, thoughtful legislation come out
of such an atmosphere?
I have looked at a number of voting systems on your site. Some of
these are quite complicated. I am guided by the following
principles:
1) For legitimacy, a majority or a near majority of the voters must
elect MLAs in individual ridings. Elected representatives must
predominate.
2) To have true debate over alternatives, minorities must be
represented and heard in the legislature. The governing party
requires real input from minority MLA'a with differing viewpoints
if it is going to enact laws in the public interest.
3) The system must be reasonably simple, convenient, and cost
effective - particularly if voters are going to accept it in the
general referendum next May.
No voting system is perfect balancing and trade-offs are
inevitable. The first priority is to reform the most damaging
aspects of the present system. In my view, the most grievous harm
is done by the first-past-the-post or plurality system when there
are more than two candidates in a riding. A distinct minority
candidate is often elected because another minority candidate
siphons votes from a stronger candidate. This is wrong in a
representative democracy. I want to support my first choice even
though that choice may have little chance to win, but I don't want
a poor third choice to win because I vote my conscience. I believe
that the 'AV/Mixed Member Proportionality' system presented in the
following sections would help to remedy this problem and the
problem of minority representation. The alternative vote system
outlined in section I would, by itself, be a significant
improvement over FPTP, simply because it would allow the voter to
freely make a first choice (even if that choice was an unelectable
minor party candidate), without worrying that this would help elect
an unacceptable alternative. This system would also tend to
increase the visibility and vote count of minor parties while
modestly increasing their representation in some cases. The
proportionality system outlined in section II would, by itself,
also be an improvement over the present system providing reasonably
proportional outcomes in most elections. I believe that a system
based on sections I and II would be a little more complicated but
an improvement over both of these systems. (For all three of these
systems you would need an additional rule to determine the
governing party. For example: the party with the most first-choice
popular votes would always have at least a one-seat plurality in
the legislature and would form the government.) However, my final
conclusion is that the best system would encompass all of sections
I, II, and III as outlined below.
I) In all the existing ridings, a majority or near-majority vote
would elect the MLAs. If there were more than two candidates, and
no candidate got a majority of the total vote, all but the two top
vote getters would be eliminated. The second choices of those
voters whose first choices were eliminated would be distributed to
the two remaining candidates to determine the winner. In this
system, a voter could vote for a first-choice candidate while
keeping a second-choice in reserve in case the first choice was
eliminated and the voting went to a second round. In most cases,
the second choices would be sufficient to determine a majority
winner. I don't think it would be a good idea to go on to third
choices if a majority winner was not attained in the second round.
The vote counting would lose its transparency and simplicity, and
the addition of third or even fourth choices to the totals
determining the winner would dilute the winner's legitimacy. Most
voters make a first choice with zero or one alternative in mind. I
believe this reform would not only make for a legislature more in
tune with actual voter preferences, both positive and negative, but
would contribute to a final result that more accurately represented
the percentage of the vote received by the top two parties. The
final legislative seat adjustment, using at-large representation,
would be made to more accurately reflect the first-choice voting
percentages received by all the parties. (See section II)
II) Assume 80 elected seats and 20 at-large seats to be determined
after the voting. Suppose the leading party got 38% of the
first-choice vote and 36 seats (45%); the second party got 36% of
the first-choice vote and 41 seats (51.25%); the third party got
12% of the first-choice vote and 2 seats (2.5%); the fourth party
got 8% of the vote and one seat (1.25%); and 6% of the vote went to
other parties/candidates. The distribution of seats is way out of
proportion to the actual voting percentages. The first party has
18.4% more than its share, (45-38)/38; the second party has 42.4%
more than its share, (51.25-36)/36; the third party has 79.2% less
than its share, (2.5-12)/12; and the fourth party has 84.4% less
than its share, (1.25-8)/8. These are the 'Representation
Advantages' or 'R' advantages of the four parties before
distribution of the 20 at-large seats. R indicates the degree of
over or under-representation and is positive or negative,
accordingly. The seat totals need to be more closely aligned to the
percentages by distributing the 20 at-large seats as follows:
a) The threshold for a party's participation in the at-large
distribution would be set at 5%.
b) The 20 at-large representatives would come from the
highest-ranking first-choice vote totals among the losing
candidates. (To compensate for differences in the voting pools
among the districts, a highest-district-percentage system could
also be used. I prefer the simplicity and transparency of absolute
vote totals.) About 20 competitive ridings would have two (rarely
more) representatives in the legislature. I believe it would be too
complicated to have the voters pick from long slates of party
candidates to determine the winners of the at-large seats. The
voters would have to consider many candidates, and the only
accountability would be to the party. In this system, an at-large
MLA would be tied to a particular district (because that's where
the votes came from), as well as the party.
c) The 20 seats would be distributed to the parties in a manner
that most closely aligned the party seat totals to the proportion
of the vote received without taking elected seats away from any
party. Also, the party with the highest percentage of the popular
vote would get at least the same number of seats as the party with
the leading elected seat total.
d) As a first approximation, the 20 at-large seats are
distributed as follows: the leading party gets 5 additional seats
(the same as the party with the leading seat total); the second
party gets no additional seats; the third party gets an additional
9 seats; and the fourth party gets the remaining 6 seats. Each
party now has a distribution of legislative seats that is more
closely proportional to the overall vote it received. The first
party has 41 seats; the second party maintains its 41 seats; the
third party has 11 seats; and the fourth party has 7 seats. Under
this distribution the first party received 7.9% more than its
proportional share, (41-38)/38; the second party received 13.9%
more than its share, (41-36)/36; the third party received 8.3% less
than its share, (11-12)/12; and the fourth party received 12.5%
less than its share. This represents the closest possible alignment
under the rules and comprises the final distribution of the 20
at-large seats. To be precise, the difference, D, between the
highest and lowest R in all possible distributions has been
minimized with D=26.4. A 40,41,11,8 distribution makes D=22.2, and
a 41,40,11,8 distribution makes D=19.4, but both of these
distributions violate rule c).
e) The above example illustrates the application of the
following rules:
i) Compute each party's proportional share of the seats based on
the first-choice voting percentages and a 100-seat legislature.
ii) Guided by the above computations, the at-large seats are
distributed to approximate each party's percentage of the votes. No
party loses any elected seats, and the party with the most popular
votes must end up with at least as many seats as the party with the
most elected seats.
iii) Compute the 'Representation Advantage', R, for each party
under the distribution. R = ((% of total seats) minus (% of total
vote)) / (% of total vote).
iv) The distribution is then adjusted, and each party's R value
is recalculated until the difference between the highest and lowest
R is minimized over all possible distributions.
f) In the above example, some might say that the first and
second parties received more than their fair share at the expense
of the third and fourth parties. But the only way to assure close
proportionality while maintaining the system of local
representation in the 80 existing ridings would be to add more
at-large seats. That would result in more losing candidates, with
lower vote totals, being included in the distribution and would
tend to dilute the local representation and legitimacy of the
legislature. It would also be more costly.
III) In the example, the two leading parties both have 41 seats
less than a 51 seat majority. Which party governs? The following
procedure would be applicable whenever an election failed to
produce a party with a majority of the seats in the
legislature:
a) Look at the second-choices of the two minor parties. Suppose
the third party's second-choices were: 40% for the first party, 50%
for the second party, and 10% for neither. The fourth party's
second-choices were: 55% for the first party, 40% for the second
party, and 5% for neither.
b) For purposes of determining the governing party, the third
party's 11 seats are 'distributed' in the ratio of 40:50 to the
first and second parties. That gives an additional 4.89 'seats' to
the first party and 6.11 'seats' to the second party. Likewise, the
fourth party's 7 seats are 'distributed' in the ratio 55:40 to the
first and second parties. That gives an additional 4.05 'seats' to
the first party and 2.95 'seats' to the second party. The total
'seats' for the first party is 41 + 4.89 + 4.05 = 49.94. The total
'seats' for the second party is 41 + 6.11 + 2.95 = 50.06. So the
second party forms the government. The rationale for this procedure
is simply that it provides a reasonably fair way to determine the
governing party by 'distributing' the minor party seats according
to the actual preferences of the minor party voters. No seats would
actually change hands after the final distribution in section II.
Thus, in a closely contested election, the second choices of the
minor party voters would play a crucial role in determining the
outcomes in individual ridings and in determining which of the
major parties actually governed. Of course, this procedure would
not make the task of governing any easier for the party that took
charge, but it would give finality and fairness to the election
process.
IV) The government would be formed by the majority or section III
minority party for the following four years. No MLA would be
permitted to switch parties during a term of office.
V) The voting system could only be changed by a provincial
referendum.
SUMMARY:
1) Election by a majority or near majority in local ridings based
on second-choice alternative voting.
2) Representation by a mixed system of local elections and
proportional distribution.
3) Government by the majority party or a minority party with
significant second-choice support.
4) Legislation enacted after real debate, negotiation, and
compromise. Better laws produced.
The strengths of this system are:
1) Election by a majority or a near majority that would better
reflect the negative as well as positive voter preferences in each
riding. In a closely-contested election, a voter could make a
'true' first-choice for a minor party candidate without fearing
that he or she would have no say in the final outcome. The parties
would get better feedback as to their actual strength among the
voters.
2) Representation in the legislature would be determined
primarily by local elections but would be better balanced by a
proportional distribution of 20 at-large seats. (20% of the total
number of seats) A closely divided electorate would be represented
by a closely divided legislature, not an artificially created
majority.
3) The governing party would always have at least its
proportional share of seats in the legislature.
4) Arrogant majority party rule would become less likely better
laws in the overall public interest would become more likely.
Government would be more open.
5) Minority governments would be more frequent. Laws would be
passed after real negotiation, debate, and compromise among the
parties. Legislators would cease being robots but would actually
serve a constructive purpose in developing and considering
legislation. Coalitions would likely form on particular pieces of
legislation.
6) Minor parties would gain a foothold in the legislature and would
have a chance to prove themselves to the voters, learn skills in
government, increase their percentage of the vote, and eventually
become the governing party.
7) The two leading parties would not be locked into their
positions by inertia and would have to prove themselves to the
voters or risk being marginalized.
8) No changes would be required in the existing ridings.
Naturally, there are questions and weaknesses in this system:
1) Is a legislature with 20% of its members 'un-elected',
legitimate? Would the minor parties wield too much power -
considering that the voters would elect very few of these parties'
representatives?
2) Would the minor parties be too focused on one or two special
issues rather than the concerns of the general public?
3) Would the vote for minor parties expand to the point of being
divisive? Would the increased likelihood of minority government
result in legislative gridlock? (In my opinion, it would be much
better to have a few bills that were thoroughly debated and
survived than many bills rammed through the legislature with no
real debate.)
4) Should the threshold for participation in the at-large seat
distribution be higher or lower than 5%? (A higher threshold could
choke off nascent, minor parties. A lower threshold could give too
much clout to numerous special-interest parties.)
5) Should the 20 at-large seats be increased or reduced?
(Increasing the number would result in more un-elected/less
representative MLA's (i.e., the MLA's would have received
relatively fewer votes in their voting districts than other
at-large MLA's) holding the balance of power on evenly-contested
pieces of legislation. Decreasing the number of seats would result
in less proportionality, more dominance by the leading party(ies),
and less-scrutinized/less-representative legislation. The above
example had a high degree of disproportionality in the initial
outcome, yet the provision for 20 at-large seats led to a final
result that was close to proportionality. In most elections it
would be possible to get reasonably close to true proportionality
with this number of at-large seats.)
6) The 20 additional seats would increase the cost of
government.
Applications of this system to the 1996 and the 2001 elections,
without the alternative voting outlined in sections I and III, are
included in the two appendices.
APPENDIX A 2001 election results under section II of this
system:
Actual results: 79 elected seats
1) Liberal: 57.6% vote 77 seats (97.5%) R= +69%
2) NDP: 21.6% vote .2 seats (2.5%) ...R= -88%
3) Green: 12.4% vote ..0 seats (0.0%) ...R= -100%
4) MP: 3.2% vote ..0 seats (0.0%) ...R= -100%
5) UP: 3.2% vote ..0 seats (0.0%) ...R= -100%
6) Other: 2.0% vote ..0 seats (0.0%) ...R= -100%
System results: 79 elected seats + 20 at-large seats (20%) = 99
seats
1) Liberal: 57.6% vote .77 seats (77.8%) R= +35.1%
2) NDP: 21.6% vote .14 seats (14.1%) R= -34.7%
3) Green: 12.4% vote . 8 seats (8.1%) ...R= -34.7%
4) MP: 3.2% vote ..0 seats (0.0%) ...R= -100%
5) UP: 3.2% vote ..0 seats (0.0%) ...R= -100%
6) Other: 2.0% vote ..0 seats (0.0%) ...R= -100%
These system results are a vast improvement over the actual results
in this extremely disproportional election. The Liberals would
still be over-represented and would dominate the legislature, but
the second and third parties would have had much more
representation. If section I alternative voting had been included,
proportionality might have further improved. For example, assuming
the NDP had elected an additional three representatives and the
Greens had elected two representatives, the final seat totals would
have been: Liberal: 72 seats (R= +26.2%), NDP: 17 seats (R=
-20.3%), and Green: 10 seats (R= -18.5%)
APPENDIX B 1996 election results under section II of this
system:
Actual results: 75 elected seats
1) Liberal: 42.0% vote .33 seats (44.0%) R= + 4.8%
2) NDP: 39.6% vote .39 seats (52.0%) R= +31.3%
3) BCR: 9.3% vote ..2 seats (2.7%) R= -71.0%
4) PDA: 5.8% vote ..1 seat (1.3%) R= -77.6%
5) Other: 3.3% vote ..0 seats (0.0%) R= -100%
System results: 75 elected seats + 19 at-large seats (20%) = 94
seats
1) Liberal: 42.0% vote .40 seats (42.6%) R= + 1.4%
2) NDP: 39.6% vote .39 seats (41.5%) .R= + 4.8%
3) BCR: 9.3% vote ...9 seats (9.6%) ....R= + 3.2%
4) PDA: 5.8% vote ...6 seats (6.4%) R= +10.3%
5) Other: 3.3% vote ...0 seats (0.0%) R= - 100%
These system results give a very close alignment with the actual
voting percentages. All the parties are marginally over-represented
at the expense of 'Other'. Note that the minimum difference rule
narrowly gives an extra seat to the PDA or BCR at the expense of
the Liberals. (A 40,39,9,6 distribution gives a difference of
10.3-1.4=8.9 between the highest and lowest R values; a 41,39,9,5
distribution gives a difference of 4.8-(-8.6) =13.4; and a
41,39,8,6 distribution gives a difference of 10.3-(-8.6)=18.9.
Also, a 41,38,9,6 distribution would have given an even lower
difference (10.3-2.0=8.3) but would have taken away an elected seat
from the NDP, and a 39,40,9,6 distribution would have given a
difference of 10.3-(-1.2)=11.5, but the popular vote leader's seat
total would not have matched the total of the party with the
leading number of elected seats.) Nevertheless, the Liberals would
have formed a minority government. I can only speculate about what
would have happened if sections I and III had been included in the
system. With respect to section I, if alternative voting had been
included, more elected seats would probably have been taken by the
Liberals than the NDP, but the overall seat totals would have been
nearly identical. With respect to section III, it is likely that
the Liberals would have obtained the necessary 7+ seats out of the
15 seats 'distributed' from the two minor parties and would have
formed the government.